J . Am. Chem. SOC.1985, 107, 7139-7145
7139
Nuclear Magnetic Resonance and Conformational Energy Characterization of Repeat Peptides of Elastin: The Cyclohexadecapeptide, cycZo-( L-Val 1-~-Pro2-G1y3-G1y4)4 M. A. Khaled,? K. U. Prasad, C. M. Venkatachalam, and D. W. Urry* Contributionfrom the Laboratory of Molecular Biophysics, School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama 35294. Received May 3, 1985
Abstract: The cyclohexadecapeptide cyclo-(L-Val,-r-Pro,-GIy3-Gly,), is synthesized and its conformational characteristics, in solution, are derived by employing ' H and I3C NMR methods. Selective decoupling and I3C ('HI NOE are used to assign the IH and I3Cresonances in particular to delineate the two glycine residues in the molecule. Temperature and solvent dependence of peptide N H chemical shift are used to evaluate secondary structure. The dominant conformational feature of this molecule is shown to be a recurring type I1 0 turn utilizing a hydrogen bond between the Gly,NH and Val, C=O groups. T , values demonstrate increased flexibility of the chain segment, Gly,-Val,, which connects the p-turns. Comparable conformational features were also obtained by using in vacuo conformational energy calculations. The conformational properties of this cyclic molecule show a close relationship to its linear counterpart. The overall conformational behavior of the tetramer series is compared with the pentamer series, ~-Val,-~-Pro~-Gly~-~-Val~-Gly~, of elastin, particularly in terms of librations within the chain segments, Va14-Gly,-Vall and Gly,-Val,, connecting the 0-turns. In analogy to the polypentapeptide study wherein a new mechanism of elasticity was developed, the present study represents a significant step in developing an understanding of the elasticity of the cross-linked polytetrapeptide of elastin.
cyclic structures had a conformation like the linear high polymer. Fibrous elastin provides the essential elasticity of such tissues The idea is that a loose helical structure with several repeats per as ligaments, lung, skin, and vascular wall. The single soluble turn could have closely the same set of conformations in one turn precursor protein of fibrous elastin is called tr~poelastin'-~ and as a cyclic molecule or, inversely, that an unstrained cyclic has been found to contain repeating sequences-a tetrapeptide (~-Val,-~-Pro,-Gly,-Gly,), a pentapeptide ( ~ - V a l , - ~ - P r o ~ - G l y , - structure containing a substantial number of residues, say ten or more, could be converted to a linear helical conformation of low L-Val,-Gly5), and a hexapeptide (~-Ala,-~-Pro~-Gly,-~-Val~pitch with only small changes in a few torsion angles.22 It was Gly,-~-Val,)."~ Extensive characterizations by physical methods have shown that the monomeric and polymeric forms of these found that the cyclopentapeptide and cyclodecapeptide had very -~ hypeptides in solution all exhibit a @ - t ~ r na, ~ten-membered different conformations from each other and from the cyclodrogen-bonded ring between the C=O of residue-1 and the N-H pentadecapeptide but that the cyclopentadecapeptide exhibited of residue-4 with Pro,-Gly, at the corners of the @-turn and a conformation very nearly identical with that of the linear poproviding the end peptide moiety with a type I1 orientation.I0." The pentapeptide and hexapeptide repeats can also exhibit a ( I ) Smith, D. w . ; Weissman, N.; Carnes, w . H. Eiochem. Eiophys. Res. y-turn, an eleven-membered hydrogen-bonded ring between the Commun. 1968, 31, 309-315. (2) Sandberg, L. B.; Weissman, N.; Smith, D. W. Biochemistry 1969, 8, Gly,NH and the G ~ Y , C = O . ~ , ~ *Under ' ~ J ~ certain conditions of 2940-2945. temperature and solvent, a 14-membered hydrogen bonded ring (3) Smith, D. W.; Brown, D. M.; Carnes, W. H. J . Eiol. Chem. 1972, 247, has also been found to O C C U ~ . * , ~ ~ , ~ ~ 2427-2432. Of particular note is that the polytetrapeptide (of interest here) (4) Gray, W. R.; Sandberg, L. B.; Foster, J. A. Nature (London) 1973, 246. 461-466. and the polypentapeptide, like tropoelastin and a-elastin16 (a (5) Foster, J. A.; Bruenger, E.; Gray, W. R.; Sandberg, L. B. J. Eiol. chemical fragmentation product of fibrous elastin), are each Chem. 1973, 248, 2876-2879. soluble in water in all proportions below 20 "C, but on raising (6) Sandberg, L. B.; Gray, W. R.; Foster, J. A.; Torres, A. R.; Alvarez, the temperature, they aggregate and, on standing, settling occurs V. L.; Janaga, J. Adu. Exp. Med. Eiol. 1977, 79, 277-284. (7) Sandberg, - L. B.; Soskel, N. T.; Leslie, J. B. N . E n d . J . Med. 1981, with formation of a dense viscoelastic phase called the c~acervate.~ 304, 566-579. The process of coacervation is reversible, and the coacervates are (8) Urry, D. W.; Long, M. M. CRC Crit. Rev. Eiochem. 1976, 4, 1-45. about half peptide or protein and half water as is hydrated fibrous (9) Urry, D. W. Meth. Enzymol. 1982, 82, 613-716. elastin. Of interest to the development of an understanding of (10) Venkatachalam, C. M. Biopolymers 1968, 6, 1425. (1 1) Khaled, M. A.; Urry, D. W. Eiochem. Eiophys. Res. Commun. 1976, the mechanism of polypeptide elasticity is that cross-linking, of 70,4a5-491. the polypentapeptide and of the polytetrapeptide of elastin either (12) Nemethy, G.; Printz, M. P. Macromolecules 1972, 6, 755-758 by chemical meansi7or by y-irradiation of the coacervate^,^^.'^ (13) Khaled, M. A.; Urry, D. W.; Okamoto, K. Eiochem. Eiophys. Res. results in elastomeric matrices with striking similarities to fibrous Commun. 1976, 72, 162-169. (14) Renugopalakrishnan, V.; Khaled, M. A,; Urry, D. W. J . Chem. SOC., elastin.20-21The similarities are with respect to elastic modulus Perkin Trans. 2 1978, 1 1 1-1 19. and temperature dependence of elastomeric force. ( 1 5 ) Khaled, M. A.; Venkatachalam, C. M.; Trapane, T. L.; Sugano, H.; The studies of conformation and elastic mechanism are wellUrry, D. W. J . Chem. SOC.,Perkin 11, 1980, 1 1 19-1 130. advanced for the polypentapeptide and the present report of the (16) Thomas, J.; Elsden, D. F.; Partridge, S . M . Nature (London) 1963, 200, 651-652. cyclohexadecapeptide of the polytetrapeptide corresponds to one (17) Urry, D. W.; Okamoto, K.; Harris, R. D.; Hendrix, C. F. Eiochemof the essential steps in the polypentapeptide study. A brief istry 1976, 15, 4083-4089. summary of the polypentapeptide studies is given for the purpose (18) Urry, D. W.; Harris, R. D.; Long, M. M. J . Eiomed. Muter. Res. of placing the present polytetrapeptide study in perspective. 1982, 16, 11-16. (19) Urry, D. W.; Prasad, K. U. In "Biocompatibility of Natural Tissues A key in developing a mechanism for the elasticity of the and Their Synthetic Analogues"; Williams, D. F., Ed.; CRC Press: Boca polypentapeptide (PPP) of elastin was the approach of examining Raton, FL, 1985; pp 89-116. the conformations of cyclic molecules containing from one to six (20) Urry, D. W. J . Protein Chem. 1984, 3, 403-436. repeating pentamer units and determining which if any of these (21) Urry, D. W.; Venkatachalam, C. M.; Wood, S . A,; Prasad, K. U. In Current address: Department of Nutrition Sciences, University of Alabama at Birmingham, Birmingham, AL 35294.
0002-7863/85/1507-7139$01.50/0
"Structure and Motion: Membranes, Nucleic Acids and Proteins"; Clementi, E., G. Corongiu, G., Sarma, M. H., Sarma, R. H., Eds.; Adenine Press: Guilderland, NY, 1985, pp 185-204. (22) Urry, D. W. Proc. Nut1 Acad. Sci. U.S.A. 1972, 69, 1610-1614.
0 1985 American Chemical Society
1140 J . Am. Chem. SOC.,Vol. 107, No. 24, 1985
lypentapeptide, and it did so in different solvents and over the accessible temperature ranges;23that is, the cyclopentadecapeptide was found to be the cyclic conformational correlate of the linear PPP. The solution conformation of the cyclopentadecapeptide was developed in detail24 (in the same manner as reported here for the cyclohexadecapeptide of the repeat tetrapeptide series), and it compared very closely to the independently determined In the crystal, crystal structure of the cyclopentade~apeptide.~~ the three type I1 Pro-Gly @-turnswere found arranged as three feet for the cyclic molecules with one cyclic molecule stacked directly on top of the other; the dominant intermolecular contacts were hydrophobic; the Val,-Gly5-Vall segment was suspended between @-turns,and the crystal was about one-third water with the water occurring within the stack of cyclic molecules. The cyclic conformation was then computationally converted to the helical linear conformational correlate26termed a @-spiral. The @-spiral is so-named because the @-turn is the dominant secondary structural feature and the pentamer conformation recurs as a helical repeat. The @-turnsfunction as spacers between the turns of the helix by means of hydrophobic contacts; there is water within the @-spiralwhich can exchange with extraspiral water through spaces in the surface of the @-spiral,and connecting the @-turns is a suspended segment, Val,-Gly5-Vall, wherein the peptide moieties have much freedom to rock, Le., to librate. These librations are viewed as a large source of entropy, and the librations are damped on extension, as would occur on stretching, such that the elastomeric restoring force would be largely entr~pic.~’This has been termed a librational entropy mechanism of ela~ticity.~,*~ Numerous physical methods have characterized the change in structural state attending formation of the viscoelastic coacervate of the polypentapeptide: (a) temperature dependence of aggregation and associated (light and transmission and scanning electron) microscopic examination of the aggregates demonstrate intermolecular ordering of polypeptide with increasing temperat ~ r e ; ~(b) , ~ nuclear ’ magnetic resonance relaxation studies demonstrate an inverse temperature transiti01-1;~~ (c) circular dichroism studies show the conversion from a more nearly random state at low temperature to a structure at 40 “C of regularly recurring @ - t u r n ~(d) ; ~dielectric ~ ~ ~ relaxation studies show the development on going from 20 to 40 “C of an intense local Debye-type relaxation at 25 MHz, indicating that each pentamer has developed the same conformation characterized by a common peptide librational m ~ d e ; ~and ~ , ~(e)’ thermoelasticity studies on the yirradiation cross-linked coacervate state show that the elastomeric force goes from essentially zero at 20 “C, before the transition, to a near maximal value at 40 “C when coacervation is complete.20,21Interestingly, the development of the intensity of the 25-MHz dielectric relaxation closely parallels the increase in elastomeric force with t e m p e r a t ~ r eas~do ~ the temperature dependences of the other physical characterizations which also indicate increase in intra- and intermolecular order with increase in temperature. Accordingly, the data on the polypentapeptide argue that a new mechanism of elasticity is required wherein the relaxed state contains a describable albeit dynamic regularity of (23) Urry, D. W.; Trapane, T. L.; Sugano, H.; Prasad, K. U. J. Am. Chem. SOC.1981, 103, 2080-2089. (24) Venkatachalam, C. M.; Khaled, M. A.; Sugano, H.; Urry, D. W . J . Am. Chem. SOC.1981, 103, 2372-2379. (25) Cook, W.J.; Einspahr, H. M.; Trapane, T. L.; Urry, D. W.; Bugg, C. E. J . Am. Chem. SOC.1980, 102, 5502-5505. (26) Venkatachalam, C. M.;Urry, D. W. Macromolecules 1981, 14, 1225-1 229. (27) Urry, D. W.; Venkatachalam, C. M.; Long, M. M.; Prasad, K. U.; In “Conformation in Biology”; Srinvasan, R., Sarma, R. H., Eds.; Adenine Press: Guilderland, NY, 1982;G.N. Ramachandran Festschrift Volume, pp 11-27. (28) Urry, D. W.Ultrastruct. Pathol. 1983, 4 , 227-251. (29) Urry, D. W.;Trapane, T. L.; Iqbal, M.; Venkatachalam, C. M.; Prasad, K. U. Biochemistry 1985, 24, 5182-5189. (30) Urry, D.W.; Prasad, K. U.; Trapane, T. L.; Iqbal, M.; Harris, R. D.; Henze, ‘American Chemical Society Division of Polymeric Materials: Science and Engineering”; Wiley: New York, 1985; Vol. 53, pp 241-245. (31) Henze, R.; Urry, D. W. J . Am. Chem. 1985, 107, 2991-2993. Henze, Rolf; Harris, R. D.; Prasad, K. U. Biochem. (32) Urry, D. W.; Biophys. Res. Commun. 1984, 125, 1082-1088.
Khaled et al.
structure as opposed to the kinetic theory of rubber elasticity wherein the relaxed state is considered to be totally random. With respect to the polytetrapeptide of elastin, the present stage is one of describing the solution conformation of a cyclic analogue that is a reasonable cyclic conformational correlate of the linear polytetrapeptide. In the present report are the synthesis and conformational characterization of the cyclohexadecapeptide, cyclo-(VPGG),. The conformational analysis of this cyclic peptide is achieved by combining NMR spectroscopy with theoretical conformational energy calculations. As only ranges of possible values for the torsion angles, 4 and $, are obtained from an analysis of the spin-spin coupling constants, these constraints are utilized in conformational energy calculations to obtain values of 4 and $ tortion angles for low-energy conformations that are consistent with the NMR data. The possible cyclic conformations contained the @-turnsand become candidates for the development of the @-spiralof the polytetrapeptide which in turn can become a rational basis with which to explore further the mechanism of elasticity of the cross-linked polytetrapeptide.
Materials and Methods Synthesis. The synthesis of cyclo-(.VPGG), was carried out by the classical solution methods. The monomeric tetrapeptide was synthesized as shown below, the details of which have been r e p ~ r t e d ’earlier. ~ Boc-Pro-OH
+ H-Gly-OBzl
TFA-H-Pro-Gly-OBzl Boc-Val-Pro-Gly-OH
mixed
anhydride
Bw-Val-OH EDC,
H-Gly-OBzI EDC,
Boc-Pro-Gly-OBzl
TFA
Boc-Val-Pro-Gly-OBzl
H2
Pd/C
Hz 3
Boc-Val-Pro-Gly-Gly-OBzl I Boc-Val-Pro-Gly-GI y-OH I1
After deblocking I with TFA (trifluoroacetic acid), it was coupled with I1 using l-(3-(dimethylamino)propyl)-3-ethylcarbcdiimide hydrochloride (EDCI)34and I-hydroxybenzotriazole (HOBt). The octapeptide benzyl ester thus obtained was deblocked and coupled with I1 to obtain the dodecapeptide benzyl ester which was deblocked and further coupled with I1 to obtain the protected hexadecapeptide benzyl ester. After hydrogenation, the peptide was converted to p-nitrophenyl ester using pnitrophenyl trifl~oroacetate.~~ After removing the tert-butyloxycarbonyl (Boc) group, cyclization was carried out at a high dilution in pyridine to obtain cyclo-(VPGG),. All the intermediate peptides were characterized by TLC (thin-layer chromatography), in different solvent systems, IH and ”C N M R spectra and elemental analyses. The T L C was performed on silica gel plates supplied by Whatman, Inc., by utilizing the following systems: R j chloroform/methanol/acetic acid (85:15:3);R j chloroform/methanol (5.1); R j n-butyl alcohol/acetic acid/water (4:l:l);Rj? chloroform/ methanol/acetic acid (75:25:3),and R j n-butyl alcohol/acetic acid/ water/pyridine (30:6:24:20).Detection of the peptides on TLC plates was with ninhydrin spray and/or chlorine-toluidine reagent. Elemental analyses were carried out by Midwest Microlab Ltd., Indianapolis, IN. Boc-amino acids were purchased from Bachem Inc., Torrence, CA, and EDCI and HOBt were purchased from Aldrich Chemical Co., Milwaukee, WI. The Pro- and Val-amino acids were of the L configuration. Boc-(Val-Pro-Gly-Gly)2-OBzl (111). To a solution of I1 (6.2g, 14.47 mmol) in D M F (35 mL) at ice-cold temperature were added HOBt (2.6 g, 17 mmol) and EDCI (3.26g, 17 mmol) and stirred for 15 min. To this was added an ice-cold solution of TFA-H-Val-Pro-Gly-GIy-OBzl (obtained by deblocking I with TFA) and triethylamine (Et,N) (2.37 mL, 17 mmol) in D M F (30 mL). The reaction mixture was stirred at 4 “ C overnight and at room temperature for 24 h. After removing DMF, the peptide was taken in CHC1, and extracted with 20% citric acid, water, saturated NaHCO, solution, and water and dried over anhydrous MgS0,. The solvent was removed under reduced pressure and the peptide was chromatographed over a silica gel column (4.1 X 52 cm) using 3% methanol in CHCI, to obtain 8.29g of the title compound: Rj, 0.76,R j , 0.63;R j , 0.75;mp shrinks at 107 OC and melts completely at 124 OC. Anal. Calcd for C,oH60N80,,: C, 57.95;H , 7.29;N , 13.51. Found: C , 58.42;H , 7.9;N, 13.13. Boc-(Val-Pro-Gly-Gly)3-OBzl (IV). After deblocking 111, the TFA salt was coupled with I1 and the dodecapeptide benzyl ester was worked (33) Urry, D. W.; Ohnishi, T. Biopolymers 1974, 13, 1223. (34) Sheehan, J. C.; Preston, J.; Cruickshank, P. A. J . Am. Chem. SOC. 1965, 87, 2492. (35) Sakakibara, S.; Inukai, N. Bull. Chem. SOC.Jpn. 1964, 37, 1231.
J. Am. Chem.Soc., Vol. 107,No. 24. 1985 7141
The Cyclohexadecapeptide up as described for the preparation of 111 (yield 98%): Rj,0.5; Rj,0.53; Rj,0.64; mp 87-89 OC dec. Anal. Calcd for CS4Hs2NI2Ol5: C, 56.92; H , 7.25; N , 14.75. Found: C, 57.10; H , 7.24; N, 14.59. Boc-(Val-Pro-Gly-Gly)4-OBzl (V). The Boc group was removed from IV and coupled with I1 to obtain V following the procedure described for I11 (yield 80%): R j , 0.46; Rj,0.32; Rj, 0.53; mp 144 OC (dec). Anal. Calcd for C68H104N16019: C, 56.33; H, 7.23; N , 15.46. Found: C, 55.91; H , 7.3; N , 15.03. Boc-(Val-Pro-Gly-Gly)4-OH (VI). Compound V (1.5 g, 1.03 mmol) in glacial HOAc (35 mL) was hydrogenated in the presence of 10% Pd/C catalyst at 40 psi. Catalyst was filtered and the solvent removed under reduced pressure. The residue was taken in saturated NaHCO? solution and extracted with ether. The aqueous solution was cooled and acidified to pH 2 with solid citric acid, and the product was extracted into CHCI,. The CHCI, extract was washed with saturated NaCl and dried and solvent removed to obtain the title compound (yield 71%): Rj,0.39; Rj, 0.08; mp 140 "cdec. Anal. Calcd for C61H98N16019: c , 53.88; H, 7.26; N , 16.48. Found: C, 54.13; H, 7.66; N , 16.04. Boc-(Val-Pro-Gly-Gly)4-ONP (VII). Compound VI (0.95 g, 0.7 mmol) in pyridine (10 mL) was reacted with p-nitrophenyl trifluoroacetate (0.66 g, 2.8 mmol), and the progress of the reaction was monitored by TLC. After the reaction was complete, pyridine was removed and the residue in CHCI, was extracted with acid and base as described for 111. Solvent was removed, and after triturating with petroleum ether, the product was obtained as an amphorous powder (yield 58%): Rj,0.25; Rj,0.23; R j , 0.51; mp 148 "C dec. Anal. Calcd for C67H101N17021:c, 54.34; H, 6.87; N , 16.08. Found: C, 54.08; H , 6.83; N , 16.37. Each of the preceding linear oligomers were examined by proton and carbon-13 nuclear magnetic resonance. By delineation of the end resonances, it was possible to verify that each tetramer addition had indeed occurred and that the product was n = 4. cyclo-(Val-Pro-Gly-Gly), (VIII). Compound VI1 was treated with TFA for 30 min, and the residue after removing TFA was triturated with ether, filtered, washed with ether, and dried. The TFA salt (0.5 g, 0.33 mmol) was taken in D M F (40 mL) and glacial acetic acid (2 mL) and added to pyridine (1.6 L) at 85-90 OC over a period of 6-8 h and stirred overnight. After removing the solvent, the residue in a mixture of methanol and water (1:l) was passed through a column of mixed-bed resin, eluting with the same solvent system. After removing the solvent, the peptide was chromatographed on an LH-20 column using methanol for elution. All the fractions showing a single spot were combined, concentrated, and lyophilized from water to give 376 mg of cyclo(VPGG), (yield 90.6%): R;, 0.41; Rj,0.58; Rj,0.64. Anal. Calcd for CS6H,8N,6016:C, 54.17; H, 7.14; N, 18.05. Found: C, 54.17; H , 7.30; N , 17.60. The molecular weight of the cyclohexadecapeptide calculates to be 1240 daltons. Since the cyclization the only significant possibilities are n = 4 and 8, vapor-pressure osmometry (Corona Wescan, Model 232A) was carried out in methanol (1.2 mg/mL) at 30 "C and the experimental estimate of molecular weight was 1350. This demonstrates that the desired cyclohexadecapeptide was obtained at greater than a mole fraction of 0.9. NMR Measurements. A 10 mM solution of cyclo-(VPGG), was made in both the C D 3 0 D and C D 3 0 H solvents for IH N M R measurements while a 30 m M solution was made in C D 3 0 D and Me2-SO-d6solvents for N M R measurements. Deuterated methanol, CD30D, was purchased from Thompson Packard, Inc., while C D 3 0 H and Me2-SO-d6 were purchased from Merck, Sharp & Dohme of Canada. All the ' H N M R spectra were obtained on a Varian HR-220 spectrometer. Proton decoupling was done by the field-sweep method using a sideband generated by a Hewlett-Packard Model 5103-A frequency synthesizer. In order to facilitate the ABX spin analysis of the glycyl methylene protons, simulated spectra were obtained with the SS-100 computer system by using a Varian Data Machine spin simulation program. 13C N M R experiments were performed on a JEOL FX-100 spectrometer equipped with a multinuclear probe operating at 25.0 MHz. A 10-ps pulse width was used for a 45" magnetization vector with a 2.5-s repetition time. Multiple selective irradiations were accomplished by using a multiirradiation accessory, JEOL Model I-MI. In all the selective decoupling experiments, the amplitude of the irradiating field, ylHHi,/2n was about 10 Hz. TI values for the 13Cnuclei were measured by using the inversion-recovery method ( 180-T-90 pulse sequence). Four to five data points were used to calculate each T I value by the leastsquares method. Conformational Energy Calculations. As discussed later, the N M R data show that within the time scale of observation, cyclo-(VPGG), exhibits 4-fold symmetry with no significant amount of cis Val-Pro bond observed. This greatly simplifies the conformational analysis. While this symmetry element markedly reduces the number of degrees of freedom of the molecule to that of the tetramer repeat unit, VPGG, a further
Prop
\
,!-
5
Li
y3
Va I ,
Figure 1. Schematic drawing of the tetramer sequence of elastin defining all torsion angles 4 , $$,and wi. Dashed arrows show representative carbon-proton interactions relevant to N M R analyses presented here. reduction by 2 of the total number of independent torsion angles occurs due to the conditions of the cyclization with the %fold symmetry.36 Figure 1 shows a schematic drawing of the tetramer unit, VPGG, where the skeletal backbone torsion angles @,, $,, and w , have been marked."sf8 When the standard bond lengths and bond angles)' and planar trans-peptide units with all w , = 180' are used, the backbone conformation of the repeat unit, VPGG, is defined by specifying the a total of eight variables. As shown values of the four pairs of bi and $i, by Go and Scheraga,'6 two torsion angles have to be adjusted to ensure exact cyclization with the 4-fold symmetry. Adjusting 4 and $ at Gly, to meet the conditions of cyclization, only six independent torsion angles are required to explore the conformational space of the molecule. Further, the proline ring closure restricts the value of & to be about -60' and the interaction between Val, and Pro2 side chains somewhat limits the range of variability of $,. Also, as discussed later, N M R analysis provides ranges of torsion angles (see Figure 6 below). The six indeqi, q52, $2, &, and $?are varied within the pendent torsion angles acceptable ranges obtained from N M R , and for each set of these angles, the cyclization conditions36are solved to obtain the corresponding values of 4, and $, leading to exact cyclization. The valyl side chain is initially fixed at the ideal "trans" conformation with x(H"-C"-Cs-HB) = 180°," and the proline ring is then fixed in the "down" y-puckered ring struct ~ r e . The ) ~ total conformational energy of the cyclic peptide is calculated for each cyclic conformation so obtained, using the potential functions of Scheraga and c o - w ~ r k e r s . ~This ~ ~ results ~~ in a set of low-energy conformations which are then refined by complete energy minimization by using Fletcher's algorithm4' with respect to the six independent torsion angles, with 4, and $4 being adjusted within the algorithm to preserve cyclization during minimization. During the later stages of the minimization iterations, the valyl side chain torsion angle xi was allowed to depart from its ideal value, corresponding to the perfectly staggered conformation about the C"-CB bond. Since the valyl side chain plays an important role in the conformational stability of this repeat peptide, calculations are repeated by using the two gauche conformations of the valyl side chains. The results are expected to delineate the role of the valyl side chain in this molecule.
Results NMR Spectral Studies. The 'H NMR resonances of the three amino acid residues, namely valine, proline, and glycine, can easily be recognized from their splitting patterns which are presented in Figure 2 . For example, both valine and proline appear as pseudotriplets in Figure 2A. On prior exchanging of the peptide N H proton, Figure 2B shows the valine a-CH proton as a doublet. However, assignment of the two glycine residues, Gly3 and Gly,, cannot be made unambiguously in this way. This laboratory has demonstrated the combined use of 'H and 13C spectra, each observed during the selective decoupling of 'H resonances, which allows simultaneous assignment of both the 'H and I3C resonances of peptide^.^^,^^ The basic principle of this technique is to selectively irradiate the a-H, and NHi+,protons, (36) Go, N.; Scheraga, H. A. Macromolecules 1973, 6, 213. (37) IUPAC-IUB Recommendations J . Mol. Bid. 1970, 52, 1. (38) Ramachandran, G. N.; Sasisekharan, V. A h . Protein Chem. 1968, 23, 283. (39) Momany, F. A.; Carruthers, L. M.; McGuire, R. F.; Scheraga, H. A. J . Phys. Chem. 1974, 78, 1595. (40) Momany, F. A,; McGuire, R. F.; Burgess, A. W.; Scheraga, H. A. J . Phys. Chem. 1975, 79, 2361. (41) Fletcher, R. Report R7125, AERE, Harwell. (42) Khaled, M. A.; Urry, D. W. J . Chem. SOC.,Chem. Commun. 1981, 230. (43) Khaled, M. A.; Harris, R. D.; Prasad, K. U.; Urry, D. W. J . Magn. Reson. 1981, 44, 255.
Khaled et al.
7142 J. Am. Chem. Soc., Vol. 107, No. 24, 1985
Table I. IH NMR Parameters of cyclo-(VPGG), in Methanol
B
residues parameters L-Val, L-Pro, Gly?" Chemical Shifts (a), ppm (fO.01) 8.55 NH 7.98 C*H 4.55 4.43 3.89,4.14 b b COH CYH 1.0, 1.04 b CbH 3.69 Coupling Constants (J), Hz (f0.25)
8.21 3.93,4.12
Temperature Coefficient, (AS/AT), ppm/OC X lo-' -6.26 -6.11 -3.13 "Values for the two glycine residues were obtained by ABX spin simulation. *Overlapped between 1.82 and 2.27 ppm. CNotanalyzed.
A
oeotide N H
a
ppm
Glya"
4 5
4 25
1
40
3 75
Figure 2. 'H NMR spectra (220 MHz) (a-proton region only) of cyclo-(VPGG), (A) in CD30H and (B) in CD30D.
across a peptide bond, which are 2J-coupled to the CFO carbon. This is shown by arrows in Figure 1. These two protons also provide almost all of the nuclear Overhauser enhancement (NOE) for the C,=O carbon. Once two such protons are located and one of the residues is identified, either from its side chain pattern or by perturbing the end groups in the case of a linear peptide, the rest of the residues in the molecule can be assigned by stepwise irradiation of the 'H resonances. Inspection of the repeat sequence in Figure 1 indicates that the Gly,NH proton is 2J-coupled to the Pro2C=0 carbon. Selective irradiation of the lowest field Gly N H proton at 8.55 ppm sharpens the lowest field C=O carbon resonance at 173.03 ppm as shown in Figure 3D (compare Figure 3C and D). Simultaneous irradiation of the same Gly N H proton and also of the Proz a - C H proton a t 4.43 ppm further reduces the line width of the same carbonyl carbon at 173.03 ppm (see Figure 3E). Note that the Val, C=O carbon at 17 1.19 ppm appears very sharp since it is devoid of the zJ coupling with an N H proton because proline, an N-substituted amino acid, is the subsequent residue. This is consistent with the previous findingd4 in cyclo-(VPGVG),. Since the C=O carbon signal at 173.03 ppm is 2J-coupled to Prop-CH, it is, therefore, assigned to the Pro2 C=O carbon. However, the line width of the Pro2 C = O carbon is not as narrow as one would expect (compare Figure 3B and E). This could be due to the 3J coupling to the Gly3CH2protons. The above-mentioned difficulty can be eliminated by measuring the selective l3C--IH NOE as indicated earlier.44 Figure 3F shows the C=O carbon resonances obtained by broad-band decoupling of all the proton resonances but without N O E (no N O E gated decoupling). Figure 3G is the result of the same experiment but with the selective saturation of the lowest field Gly, NH proton a t 8.55 ppm. The difference spectrum, depicted in Figure 3H, exhibits only the Pro2 C=O carbon resonance a t 173.03 ppm. Selective heteronuclear NOE, therefore, seems to be a convenient method for the 'H and 13Csignal assignments. Since the Gly N H proton signal at 8.55 ppm is, thus, assigned to the Gly, residue, the other Gly "proton at 8.21 ppm, by elimination, is, therefore, assigned to the Gly, residue. Selective irradiation of this Gly, NH proton delineated the Gly, C=O carbon at 170.12 ppm from (44) Khaled, M. A.; Watkins, C. L. J . Am. Chem. SOC.1983, 205, 3363.
the Gly, C=O carbon resonance at 169.49 ppm since the Gly, N H proton is 2J-coupled to the Gly, C=O carbon (see Figure 1). Having made all the assignments, the 'H N M R parameters so obtained are listed in Table I and 13C N M R parameters in Tables I1 and 111. The first step in the structural evaluation of a peptide is to find out if there is any hydrogen bond formed within the molecule. Generally, two 'H N M R methods are used to detect H-bonded (solvent-shielded)peptide NH protons. These are the temperature dependence of the peptide N H protons and the solvent dependence (from basic to protic solvents) of the chemical shifts of the peptide NH proton~.''~,~~ The peptide N H proton chemical shift variation, A6, is monitored as a function of temperature. The numerical values of temperature coefficient (A6/AT) obtained in MeOH are included in Table I, and the graphical representations are given in Figure 4. The peptide N H proton solvent shift (As/AS) from the weakly basic solvent, Me2SO-d6,to the protic solvent, trifluoroethanol (TFE), was followed. In order to relate the structural behavior in MeOH, a similar solvent titration from MeOH to TFE was also followed. These are presented in Figure 5. I3C N M R has also been utilized to delineate the intramolecularly H-bonded peptide C=O groups.8 The chemical shifts of an exposed C=O carbon vary markedly on going from a basic solvent, e.g., MezSO-d6 to a protic solvent, e.g., water. Solvent titrations of the carbonyl carbon resonances of cyclo-(VPGG), were, therefore, performed, and the results are summarized in Table 11. The solvent shifts are listed both as the change in chemical shift with change of solvent (A6/ASl,) and as the change in chemical shift AI?^^) calculated with respect to the carbonyl carbon resonance which showed the lowest downfield shift. Table I1 also contains the similar 13C N M R parameters for the linear polytetrapeptide, (VPGG),, for the purpose of comparing the spectral behavior between the cyclic and the linear (VPGG) fragments. It can be noticed in Figure 2 that the Gly, C H , protons are more nonequivalent than the Gly, CH2 protons. This may be due to the increased flexibility at the Gly4 position. Spin-lattice relaxation time ( T I ) studies were carried out to access such flexibility in ~ y c l o - ( v P G G ) ~TI . values obtained in CD,OD are given in Table 111. Conformational Energy Calculations. When the six independent torsion angles are varied within the acceptable ranges obtained from N M R (Figure 6), successful cyclization resulted in several hundreds of sets of angles. About 2000 cyclic structures were obtained. Energy calculations with a trans valyl side chain gave several low-energy conformations. Most of the resulting torsion (45) Kopple, K. D.; Ohnishi, M.; Go, A. J . Am. Chem. SOC.1969, 92, 4264. (46) Ohnishi, M.; Urry, D. W. Biochem. Biophys. Res. Commun. 1969, 36, 194.
J. Am. Chem. SOC., Vol. 107, No. 24, 1985 1143
The Cyclohexadecapeptide
F
c
I 85
90
75
BO
vpm
4 5
4 25
4 0
3 75
Figure 3. Selective irradiation experiment: (A) 220-MHz 'H spectrum of cycl~-(vPGG)~ in CH,OH (a-and N H proton regions only); (B-H) 25-MHz "C NMR of C=O carbons of cyclc-(VPGG) in CD,OH. (B)broad-band IH irradiation with NOE, (C) IH undecoupled, (D) Gly, NH proton selectively irradiated, (E) both Gly, N H and Pro, C"H protons selectively decoupled, (F) broad band proton decoupled without NOE, (G) same as F with NOE, and (H) difference spectrum obtained by subtracting F from G. Note that the irradiated resonances are indicated by the filled triangles (A),while the observed signals are indicated by the open triangles (A). Table 11. Solvent Shifts (A6/AS)' of the Carbonyl Carbon Resonances cyclo-(VPGG), H-(VPGG),-Val-OMe residue Me,SO-d, D20 A6/M A6mlb Me,SO-d, DZO A6/AS A6mIb 1.66 0 171.04 172.84 1.80 0 171.18 172.84 Val,C=O 2.57 0.9 1 172.99 175.56 2.57 0.77 173.03 174.60 Pro2C=0 2.72 1.06 170.02 172.69 2.67 0.87 170.12 172.84 GIy3 C-0 0.57 169.44 171.67 169.49 171.72 2.23 2.23 0.43 Gly,C=O oSolvent shift (Aa/AS) from Me2S0 to D20, expressed in ppm. bA6m, = (A6/ASs,), - (A6/ASJI where m stands for the mth residue and 1 for residue 1; i.e., the residue 1 carbonyl carbon resonance is used as an internal reference since it exhibited the least solvent dependence. Table 111. 13C NMR Parameters of cyclo-(VPGG), in Methanol chem shifts, spin-lattice residue C atoms PPm relaxation time (TI),s Val, c" 62.27 0.174 d 3 1.79 0.207 0 19.66, 18.78 1.288, 1.814 C=O
Pro,
c" 0 c7
cb GlY, GlY4
C=O c* C=O ?I
C=O
172.50 58.05 30.34 26.31 a 174.93 43.29 172.11 43.78 171.38
2.589 0.161 0.267 0.399 a 2.603 0.190 2.305 0.208 2.179
"Overlapped with the solvent peaks. angles are found to be generally within or near the acceptable ranges. The values of 4 and $ a t Gly, were found to show con-
siderable variation among the low-energy structures. The lowest energy structure exhibits a trans valyl side chain. Higher conformational energies are obtained if the valine side chain is fixed in the gauche conformations. The torsion angles of six low-energy structures are listed in Table IV, and Figure 7 depicts their stereoviews.
Discussion The appearance of a single NMR resonance for a particular nucleus or group of equivalent nuclei would be indicative either of a single, well-defined conformation of the molecule or of a set of rapidly interconverting structures. It can be noticed in Figure 3, particularly in the NH proton and the C=O carbon regions, that there are additional detectable weak signals. These additional signals could be due to either a cis-trans isomerism at the Val,-Pro, , ~ due ' to a peptide bond, as was observed in ~ y c l o - ( V P G ) ~or conformation which does not maintain a 4-fold symmetry for (47) Khaled, M. A,; Renugopalakrishnan, V.; Sugano, H.; Rapaka, R. S.; Urry, D. W. J . Phys. Chem. 1978,82, 2743.
Khaled et al.
1144 J . Am. Chem. Soc., Vol. 107, No. 24, 1985 Table IV. Torsion Angles of the Low Structures of cyclo-(VPGG), Val, Proz struct A
B C D E F
GlY3
d
*
@
*
9
-86 -90 -87 -1 14 -150 -60
122 120 126 88 121 150
-60 -60 -60 -60 -60 -60
140 122 125 155 138 160
138 121 142 99 101 138
A
*
d
*
E , kcal/mol
-38 35 -36 37 -39 -3 7
55 -8 7 63 -1 10 81 -68
87 180 111 -1 32 -177 120
21.6 22.7 23.8 23.9 27.2 28.3
GlY4
B AI
._ L v)
80
8.5
-
, 20
0
40
EO T e m p e r a t u r e ("C)
Figure 4. Temperature dependence of peptide NH protons of cyclo(VPGG),: (A) in CD30H and (B) in TFE. A
-1
B
71
4-
60
80
/ I
I O ~ L
20
40
60
80
Figure 6. Possible ranges of the backbone torsion angles consistent with the observed NMR coupling constants depicted here by using a polar representation for the torsion angles. The solid regions refer to the torsion angles q5 and the hatched regions to the angles IC. Newman projections for typical angular values are also shown to facilitate the examination of these ranges and to make apparent the convention used.
I
for
% Triflouroethonol i v h )
Figure 5. Solvent shifts of peptide N H protons of cyclo-(VPGG),: (A) from Me2SO-d4to T F E and (B) from CD30H to TFE.
cyclo-(VPGG), in solution. We are, however, concerned with the major form of the molecule giving the most intense signals (see Figure 2) which indicate that this form of the molecule retains the 4-fold symmetry in solution within the 220-MHz NMR time scale. When a peptide N H proton is involved in an energetically most favorable interaction with the C=O oxygen and/or in a close contact with any other atoms within the molecule, it shows lowtemperature dependence of its chemical shift. From the temperature studies of the peptide N H proton chemical shifts of cyclo-(VPGG), in M e O H and TFE, as shown in Figure 4, it appears that the Gly4NH is least shifted. The temperature coefficients (AS/AT), tabulated in Table I, also show that the Gly, N H proton gives a A&/ATof -3.13 X lo-, ppm/OC, while the Val, N H and Gly, N H protons give values of -6.26 and 6.1 1 X lo-, ppm/OC, respectively. This indicates that the Gly4NH group is well-shielded while Val, and Gly, N H groups are highly exposed.8 It can be seen in Figure 5 that both Val, and Gly, N H protons show upfield shifts, whereas the Gly, N H proton does not show
Figure 7. Stereoplots of the low-energy structures obtained from conformational energy calculations. The torsion angles and energies of these structures are given in Table IV.
any appreciable chemical shift change on going from Me,SO-d, to T F E (Figure 5A) and from MeOH to T F E (Figure 5B). In the Gramicidin S model system, the exposed N H proton gave an upfield shift of about 1 ppm by the variation of solvents.48 On the basis of this criterion, the Val, and Gly, NH groups are, therefore, solvent-exposed, while the Gly, N H is solvent-shielded, Comparison of the shielding of the peptide N H group with that of the peptide C=O group provides information on the H-bond pair, Le., the NH-.O=C pairing. In the I3C N M R experiments, carbons obtained in the the solvent shifts of peptide -0 Me2SO/water solvent are listed in Table 11. It can be seen that the Val, C=O carbon is least-shifted, 1.66 ppm, compared to all other peptide C==O carbons. This means that the Val, C=O is most involved in intramolecularly H-bonding interaction. In(48) Pitner, T. P.; Urry, D. W. J . Am. Chem. SOC.1972, 94, 1399.
The Cyclohexadecapeptide terestingly, the A6/AS values of peptide C=O carbons in cyclo-(VPGG), are very similar, within the experimental error, to those of linear polytetrapeptide (VPGG), (see Table 11). As mentioned earlier, the linear VPGG molecule gave rise to a 10membered H bond, @-turn (type 11), between the Val, C=O and the Gly, N H groups. It is also apparent here in cyclo-(VPGG), that the same @-turn,with Pro2-Gly, forming the end peptide moiety, is the main conformational feature. Further conformation of this is that the cyclohexadecapeptide and the linear polytetrapeptide exhibit similar P-turn C D patterns.49 The conformational torsion angle, 4, can be estimated from the observed Coupling COnStantS.sO Each , J N H + ~Value H can give four possible ranges of the 4 torsion angles. The ranges of the 4 torsion angles for Val,, Gly,, and Gly,, obtained from their 3JNH-CaH coupling values (see Table I), are presented in Figure 6. The ranges of the torsion angles, I),can be obtained from the observed 2 J ( H Hcoupling ) for the two glycine residues only.s' They are also shown in Figure 6. Measurement of NOE between the CaH, and the NHi+, protons across a peptide can also provide information on the torsion angle, $, for residue i.11,52For example, the observed NOE between the Pro2 C*H and the Gly, N H protons could provide the value for I)2 of Pro2. Unfortunately, such experiments could not be performed at 100 M H z because the intense solvent CD,OH proton signals mask the a-IH region. For a cyclic molecule of this size with only one hydrogen bond per repeat unit, conformational flexibility is expected. This is evident from examination of wire models and from the conformational energy calculations. This molecular flexibility is also apparent from the TI values of all the a - C atoms. An identical TI value for all the a-carbons is expected to be seen for a rigid molecule which rotates isotropically in solution.53 The T I values for cyclo-(VPGG), (see Table 111) show that the Val, a-C carbon has a higher T I (0.174 s) than the Pro2 a - C ( T I = 0.161 s), and similarly the Gly, a-C carbon T I (0.208 s) is higher than that of the Gly, a-Ccarbon (0.190 s). It is appreciated that glycine with two a-protons usually gives higher T I values, but it appears that Gly, is more flexible than Gly3. This is also consistent with the observed smaller nonequivalency of its methylene (CH,) protons (see Figure 2 ) . The 3JNH-c.Hcoupling value of 8.0 Hz for the Val, residue also allows for some rotation at the N-C" bond of this residue which is consistent with its higher TI value compared to T , of Pro2 a-C carbon (see Table 111). A reasonable explanation of these observations is that cyclo-(VPGG), contains the @-turn (type 11) placing the Pro2-Gly, fragment within the turn and thus rendering these two residues more rigid. On the other hand, the Val, and Gly, residues are much more flexible, giving rise to appreciable librational freedom for the Gly,-Val, peptide unit. The flexibility of this molecule is also demonstrated by the existence of several minimum energy structures that are found from the conformational energy calculations. Many of these (49) Urry, D. W.; Long, M. M.; Ohnishi, T.; Jacobs, M. Biochem. Biophys. Res. Commun. 1974, 61, 1427-1433. (50) Bystrov, V. F.; Portnova, S. L.; Tsetlin, V. I.; Ivanov, V. T.; Ovchinnikov, Yu A. Tetrahedron 1969, 25, 439. (51) Barfield, M.; Hruby, V. J.; Meraldi, J. P. J . A m . Chem. S o t . 1976, 98, 1308. (52) Roques, B. P.; Rao, R.; Marion, D. Biochimie 1980, 62, 753. (53) Allerhand, A.; Komoroski, R. A. J . Am. Chem. SOC.1973, 95, 8228.
J . Am. Chem. SOC., Vol. 107, No. 24, 1985 7145 low-energy structures are similar to each other, and hence it suffices to show a handful of representative structures. These are given in Figure 7. Comparison of the torsion angles of these structures with the NMR-derived ranges in Figure 6 shows satisfactory agreement in general. Some discrepancies may be seen and 4,. The absence of an particularly in the values of 4,, exact agreement between the torsion angles of the low-energy structures with all the N M R ranges suggests that conformational averaging is occurring in this molecule. The structural analysis is, therefore, that the cyclohexadecapeptide contains four repeating @-turnsand that the chain segments connecting the @-turnsare highly flexible. The result for this cyclic conformational correlate of the polytetrapeptide is quite analogous to that of the cy~lopentadecapeptide~~ which is the cyclic conformational correlate of the p~lypentapeptide~~ and which is the basis for the development of the dynamic @-spiralof the polypentapeptide.26 Several polypentapeptide @-spiralsare considered to supercoil to form the 5.5-nm twisted filaments observed in the transmission electron micro~cope.~'The heat-elicited aggregation of the polytetrapeptide results in the observation (in the transmission electron microscope of negatively stained aggregates) of twisted filaments with a width of about 5 nm7.54 The twisted filaments are taken as the fundamental structural unit within the viscoelastic coacervate of the polytetrapeptide, and it is y-irradiation cross-linking of the polytetrapeptide coacervate that results in the formation of an elastomeric matrix.I8 As a @-spiralstructure forms on raising the temperature in water,s4one looks to intramolecular hydrophobic association, optimized as interturn hydrophobic interactions, to be the structural force leading to @-spiral formation. Also, as the temperature required for coacervation decreases with increasing polytetrapeptide concentration, hydrophobic intermolecular interactions are the dominant process in the formation of the twisted filaments from association of @-spirals. These intramolecular and intermolecular hydrophobic interactions become the guiding principles for utilizing the conformations in Figure 7 to develop the @-spiralof the polytetrapeptide and to develop the twisted filaments. With respect to the mechanism of elasticity, the flexible chain segment connecting the @-turns and involving the sequence of torsion angles-4,, I),,4,--becomes the focal point for the source of librational entropy for the polytetrapeptide @-spiral. The flexibility and librational entropies, in analogy to the @-spiralof the p~lypentapeptide,~',~~ are expected to decrease on extension of the polytetrapeptide @-spiral. Entropy is then expected to provide a significant component of the elastomeric force of the cross-linked polytetrapeptide.
+,,
Acknowledgment. This work was supported in part by the National Institutes of Health, Grant HL-29578. Registry No. I, 79252-56-5; 11, 53488-00-9; 111, 98509-19-4; IV, 98509-20-7; V, 98509-21-8; VI, 53488-05-4; VII, 98509-22-9; VIII, 53533-24-7; H-Val-Pro-Gly-Gly-OBzl-TFA,98509-24-1; H-(Val-ProGly-Gly),-OBzl.TFA, 98509-26-3; H-(Val-Pro-Gly-Gly),-ONP.TFA, 98509-28-5. (54) Long, M. M.; Rapaka, R. S.; Volpin, D.; Pasquali-Ronchetti, I.; LJrry,
D.W. Arch. Biochem. Biophys. 1980, 201, 445-452.
